RESEARCH ARTICLE

MicroRNA-134 Contributes to Glucose-Induced Endothelial Cell Dysfunction andThis Effect Can Be Reversed by Far-InfraredIrradiationHsei-Wei Wang1,2,3☯‡, Shu-Han Su1,4☯‡, Yen-Li Wang1, Shih-Ting Chang5, Ko-Hsun Liao1,Hung-Hao Lo1, Ya-Lin Chiu1, Tsung-Han Hsieh1, Tse-Shun Huang6, Chin-Sheng Lin7, Shu-Meng Cheng7, Cheng-Chung Cheng7*

1 Institute of Microbiology and Immunology, National Yang-Ming University, Taipei, Taiwan, 2 VGH-YMGenome Research Center, National Yang-Ming University, Taipei, Taiwan, 3 Department of Education andResearch, Taipei City Hospital, Taipei, Taiwan, 4 Program in Molecular Medicine, National Yang-MingUniversity and Academia Sinica, Taipei, Taiwan, 5 Institute of Biomedical Informatics, National Yang-MingUniversity, Taipei, Taiwan, 6 Institute of Engineering in Medicine, University of California, San Diego, LaJolla, California, United States of America, 7 Division of Cardiology, Department of Internal Medicine,Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan

☯ These authors contributed equally to this work.‡HWWand SHS are first authors on this work.* chengcc@mail.ndmctsgh.edu.tw

AbstractDiabetes mellitus (DM) is a metabolic disease that is increasing worldwide. Furthermore, it

is associated with the deregulation of vascular-related functions, which can develop into

major complications among DM patients. Endothelial colony forming cells (ECFCs) have

the potential to bring about medical repairs because of their post-natal angiogenic activities;

however, such activities are impaired by high glucose- (HG) and the DM-associated con-

ditions. Far-infrared radiation (FIR) transfers energy as heat that is perceived by the thermore-

ceptors in human skin. Several studies have revealed that FIR improves vascular endothelial

functioning and boost angiogenesis. FIR has been used as anti-inflammatory therapy and as

a clinical treatment for peripheral circulation improvement. In addition to vascular repair, there

is increasing evidence to show that FIR can be applied to a variety of diseases, including car-

diovascular disorders, hypertension and arthritis. Yet mechanism of action of FIR and the bio-

markers that indicate FIR effects remain unclear. MicroRNA-134 (miR-134-5p) was identified

by small RNA sequencing as being increased in high glucose (HG) treated dfECFCs (HG-

dfECFCs). Highly expressed miR-134 was also validated in dmECFCs by RT-qPCR and it is

associated with impaired angiogenic activities of ECFCs. The functioning of ECFCs is

improved by FIR treatment and this occurs via a reduction in the level of miR-134 and an

increase in the NRIP1 transcript, a direct target of miR-134. Using a mouse ischemic hindlimb

model, the recovery of impaired blood flow in the presence of HG-dfECFCs was improved by

FIR pretreatment and this enhanced functionality was decreased when there was miR-134

overexpression in the FIR pretreated HG-dfECFCs. In conclusion, our results reveal that the

PLOSONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 1 / 27

OPEN ACCESS

Citation:Wang H-W, Su S-H, Wang Y-L, Chang S-T,Liao K-H, Lo H-H, et al. (2016) MicroRNA-134Contributes to Glucose-Induced Endothelial CellDysfunction and This Effect Can Be Reversed by Far-Infrared Irradiation. PLoS ONE 11(1): e0147067.doi:10.1371/journal.pone.0147067

Editor: Zheng Li, Peking Union Medical CollegeHospital, CHINA

Received: August 24, 2015

Accepted: December 27, 2015

Published: January 22, 2016

Copyright: © 2016 Wang et al. This is an openaccess article distributed under the terms of theCreative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in anymedium, provided the original author and source arecredited.

Data Availability Statement: All relevant data arewithin the paper and its Supporting Information files.

Funding: Provide materials: Tri-Service GeneralHospital [TSGH-C103-023, TSGH c104-028] Cheng-Chung Cheng received these fundings. Provideregents, human resources: Ministry of Science andTechnology [MOST; 103-2321-B-010-026, 103-2911-I-010-506, and 104-2911-I-010-506 and 104-2321-B-010-027], Veterans General Hospitals UniversitySystem of Taiwan (VGHUST) Joint ResearchProgram, Tsou’s Foundation [VGHUST103-G7-2-3],National Health Research Institutes [NHRI-EX103-

deregulation of miR-134 is involved in angiogenic defects found in DM patients. FIR treatment

improves the angiogenic activity of HG-dfECFCs and dmECFCs and FIR has potential as a

treatment for DM. Detection of miR-134 expression in FIR-treated ECFCs should help us to

explore further the effectiveness of FIR therapy.

IntroductionDiabetes mellitus (DM) is one of a group of chronic metabolic disorders that are characterized byinappropriate hyperglycemia and it has been estimated that the disease will affect 7.8% of theworld adult population by 2030 [1]. The predominant form of diabetes is type 2 (T2D), which isaccounts for nearly to 90% of cases, with the other 10% consisting of type 1 diabetes and gesta-tional diabetes (GDM). In addition to hyperglycemia, the presence of high glucose impairedblood vessels and aberrant angiogenesis that contribute to many of the clinical manifestations ofdiabetes and are important caused of mortality among DM patients [2]. Thus it is important toidentify protective approaches and therapeutic strategies that minimize the complications of DM.

Endothelial colony forming cells (ECFCs, also called late outgrowth endothelial cells, OECs)are circulating endothelial cells that express endothelial lineage surface antigens such as plate-let/endothelial cell adhesion molecule 1 (PECAM1), cadherin 5 type 2 (vascular endothelium)(CDH5, also called VE-cadherin), kinase inset domain receptor (KDR, also called VEGFR2)and hematopoietic stem cell marker (CD34), but lack hematopoietic lineage markers (CD14and CD45) [3, 4]. ECFCs are potential tissue engineering materials because they show signifi-cant proliferation and de novo tubulogenic ability [5]. ECFCs are capable of being incorporatedinto the resident vasculature directly and consequently help recovery of damaged vascularregions in in vivo ischemia models [3, 6–8]. ECFCs after high glucose treatment or isolatedfrom GDM pregnancies show a slower cell proliferation rate, impaired cell migration abilityand poorer tube formation ability [9, 10]. An investigation into the molecular mechanisms atwork in ECFCs from a diabetic environment or when grown in high glucose conditions shouldaid our understanding of how to bring about therapeutic improvements in angiogenesis.

MicroRNAs (miRNAs) are small non-coding RNA molecules that are 21~23 nucleotides inlength and are crucial for posttranscriptional gene regulation [11]. miRNAs mediate a widerange of cellular processes by inhibiting their targets through either translational repression ormRNA degradation. Dysregulation of miRNAs has been observed to be associated with a rangeof human pathologies including cancer, neurodegeneration and vascular diseases [12–15].miRNAs are able to target pro-angiogenic or anti-angiogenic factors that play critical roles incontrolling angiogenesis. In DM, several miRNAs have been identified as being involved in theregulation of pancreatic beta-cells; they do this by modulating cell growth, by controlling insu-lin biosynthesis and secretion, or by targeting tissues on which insulin acts, including liver,muscles and adipocytes [16]. Microvascular and macrovascular complications, which are char-acteristics of blood vessel damage and ischemic events, are the major complications in diabeticpatients. miRNAs are dysregulated during the process of tissue hypoxia and manipulatingthese miRNAs improves postischemic revascularization. In mouse myocardial infarctionmodel, intramyocardially injection of miR-210 precursor improves vascularization [17]. miR-100 is downregulated after femoral artery occlusion and inhibition of miR-100 restores perfu-sion in hindlimb ischemic region in vivo [18]. In this study, we evaluate the effect of miRNA inECFC mediated restoration of blood flow perfusion using mouse hindlimb ischemia model.However, the regulatory mechanisms of miRNAs within DM and the consequent effects toECFCs remain largely unclear.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 2 / 27

10254SI], Taipei City Hospital [10401-62-014],National Science Council [102-2911-I-010-503, 101-2320-B-010-059-MY3] and National Yang-MingUniversity [Ministry of Education, Aim for the TopUniversity Plan]. This work was also supported in partby the UST-UCSD International Center for Excellencein Advanced Bioengineering sponsored by theTaiwan MOST I-RiCE Program [103-2911-I-009-101].Hsei-Wei Wang received these fundings.

Competing Interests: The authors have declaredthat no competing interests exist.

Far infrared radiation (FIR) is an invisible form of electromagnetic energy that has a wave-length that is longer than visible light. There are three main techniques can be used for FIRradiation delivery, namely FIR saunas, FIR ray devices and FIR emitting ceramics and fabrics[19]. FIR is a promising treatment strategy for a number of medical conditions, such asimprovement of sleep [20], inhibition of chronic pain [21]. FIR has also been shown to improveendothelial function, peripheral arterial disease and coronary artery-related ischemia [22–26].In addition, FIR treatment promotes blood flow recovery after ischemia surgery in streptozoto-cin-induced diabetic mice. However, whether FIR improves the functional properties of ECFCamong DM patients by allowing the repair of ischemia region and whether miRNAs contributeto FIR mediated restoration of blood flow remain largely unknown.

In this study, we investigated the role of miRNAs in impairing the functional abilities ofECFC under DM and high glucose (HG) condition and the decreased angiogenic activities canbe reversed by FIR. Our results provide insights into the miRNA regulation that underlies theFIR treatment of HG-dfECFCs, and also reveals a new clue that how the FIR effect promotesHG-dfECFC activity during recovery from ischemic situation.

Materials and Methods

Ethics statementThis study was approved by the Institutional Ethics Committee/Institutional Review Board ofTaipei City Hospital Heping Fuyou Branch (IRB number: TCHIRB-1030507) and Tri-ServiceGeneral Hospital (IRB number: 1-103-05-061). Patients diagnosed with Type II DM areincluded in this study. All participants provide their written informed consent to participate inthis study. The protocols of this study are consistent with the ethical guidelines of the 2008 Hel-sinki Declaration. Animal experimental procedures were approved by the Institutional AnimalCare and Use Committee (IACUC) of National Yang-Ming University and are in compliancewith the ARRIVE guidelines.

Isolation of peripheral blood ECFCs from DM patients and disease-freedonorsECFCs were collected from the peripheral blood samples of disease-free (DF) subjects or DMpatients and were isolated by density centrifugation using Histopaque-1077 (1.077 g/ml, Sigma,St. Louis, MO, USA). Isolated mononuclear cells (MNCs) were cultured in fibronectin-coatedplates with Endothelial Growth Medium-2 (EGM-2; Lonza Ltd) containing complete supplement(Hydrocortisone, IGF-1, hEGF, hVEGF, hFGF2, ascorbic acid, GA-1000, heparin and 2% FBS).After 4–5 days of cultivation, the unattached cells were removed and the culture medium wasrefreshed. At 2–3 weeks, the attached ECFCs had a cobblestone-shaped appearance with a mono-layer growth pattern. These ECFCs expressed both endothelial and hematopoietic stem cell sur-face markers as described earlier [3, 4, 27]. The ECFCs used in our experiments were frompassage 2 to passage 6 in order to avoid any aging or senescence issues. In order to imitate thephysiological DM condition, ECFCs were treated with 0.5M D-glucose stock solution in EGM-2to the final concentration at 25 mM to form the high glucose (HG) group, which mimics hyper-glycemia. Similarly, ECFCs were cultured in EBM-2 containing 2% FBS to form the low growthfactor (LGF) group. In addition, ECFCs were cultured in LGF combined with 25 mMD-glucose(HG/LGF) to mimic advanced DMwhen hyperglycemia is accompanied by ischemia-inducedtissue starvation. Finally, ECFCs were treated with D-mannitol to form the osmotic controlgroup. The concentration of 25 mMD-glucose used for treatment is equivalent to the 450 mg/dlin human blood and this mimics that situation in type II DM patients [28].

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 3 / 27

Isolation of human umbilical vein endothelial cells (HUVECs)The umbilical cord vein perfused and digested with collagenase for 30 minutes at 37°C.DMEM with 10% FBS was used to drain out the endothelial cells from the lumen of the veinand these were then centrifuged. The cells were cultured in fibronectin-coated flasks withM199 medium containing 20% FBS, 5 μg/ml heparin, and 100 μg/ml ECGs. After 6 hours, theHUVECs had attached and the medium was replaced by EGM-2 (Lonza Ltd).

Quantitative RT-PCRRNA from cell lysates and plasma were extracted using Trizol1 and Trizol1 LS reagent (LifeTechnologies) respectively, according to the manufacturer's instructions [29]. For detection ofmiRNA expression, the expression levels of specific miRNAs were detected using the stem-loop strategy described in a previous study [30]. The specific primers for each miRNAs arelisted in S1 Table. In order to measure gene expression, cDNA was synthesized using a Rever-tAid First Strand cDNA Synthesis Kit (Thermo Scientific). All the specific products of the miR-NAs and genes were detected using Maxima™ SYBR Green qPCRMaster Mix (ThermoScientific) on a StepOne™ sequence detector (Applied Biosystems, USA). The miRNA expres-sion levels from the ECFCs and plasma were normalized against U48 and miR-16, respectively[27, 31]. The mRNA expression levels were normalized against average GAPDH level.

Gene and miRNA overexpression/knockdownIn order to overexpress various miRNAs, the precursor sequences of the miRNAs were clonedinto lentiviral expression plasmids using the primer pairs listed in S2 Table. For NRIP1 knock-down, the shRNA was obtained from the National RNAi Core Facility in Academia Sinica, Tai-wan. The plasmids for miRNA overexpression and NRIP1 knockdown were packaged intolentiviral particles using 293T cells and the protocol is briefly described below. A total of 2 μgpMDG, 2 μg pCMVDR8.91 and 3 μg Lenti-miRNA or shRNA plasmid were cotransfected intohuman embryonic kidneys (HEK)-293T cells using 14 μl Turbofect (Thermo Scientific™ Fermen-tas™, USA) regent. After 48 hours, the supernatant was collected and preserved at -80°C until fur-ther use. ECFCs seeded in 6-well plate with 80% confluency were used for supernatant infectionand media were replaced with EGM2 containing complete supplements after 6 hours. The over-expression or knockdown efficiencies were evaluated 48 hrs post infection using qPCR.

To knock down miR-134 in ECFCs, commercial synthetic antagomir with chemical modifi-cation (2'-OMe-RNA backbone, first two and last four bases phosphorothioated, 3’-cholesteroltail) (RiboBio Co., Guangzhou, China) were added into culture medium at a final concentra-tion of 50 nM at 70~80% cell confluence. The expression level of miR-134 was measured byqPCR after 48 hours.

Transwell cell migration assay, tube formation assay, cell proliferationassay and cell cycle analysisFor the Transwell migration assay, 600 μl EGM2 medium with 10% FBS was added to thelower chamber, while 5 x 104 ECFCs in 100 ul EGM2 medium was seeded into the upper 8 μmPolycarbonate Permeable Transwell chamber; the unit was then incubated at 37°C. After 6hours, the supernatant was removed from the upper chamber and the 8μmmembranes fixedwith 4% paraformaldehyde for 15 minutes at room temperature. Hoechst 33342 reagent(Sigma-Aldrich) was added for 30 minutes to carry out nuclear staining. Migrated cells werecounted under fluorescent microscopy across five representative fields.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 4 / 27

For the in vitro tube formation assay, 50 μl thawed Basement Membrane Extract (TrevigenInc.) was coated on 96-well plates at 37°C for 1 hour to form a reconstituted basement mem-brane. Next 7.5 x 103 cells in 100 μl EGM-2 were seeded on the Matrigel and incubated at 37°Cfor 3–6 hours. Tube structures were then inspected under an inverted light microscope (10 X)and five representative fields in each group were captured in order to calculate total tubelength. The total tube lengths were analyzed by ImageJ and are presented as relative tube lengthafter normalization against control group.

For the MTT cell proliferation assay, 1 x 104 cells in EGM-2 per well were cultured in fibro-nectin coated 24 well plates. The cell number was evaluated from day 0 to day 6. 450 μl MTTreagent (Invitrogen, USA) was added and incubated at 37°C. After 2.5 hours, MTT reagent wasremoved, 500 μl of 0.1% sodium dodecyl sulfate (SDS)-isopropanol solution was added and theformazan was separated out. A total of 200 μl solution from each well was transferred to96-well plate and was determined using ELISA reader at 570 nm and 650 nm.

For cell cycle analysis, 1 x 106 cells were fixed in cold 70% alcohol at -20°C overnight. Next0.1% TritonX-100 containing 0.2 mg/ml RNase was used to permeabilize the cells for 1.5 hoursat 37°C. Cells were then washed using PBS and stained with propidium iodide (PI) for 30 min-utes. The cell cycle analysis was then carried out on a FACACanto (BD Pharmingen) and cal-culated by FlowJo.

Each experiment was examined with at least three unique patient lines and two replicatesper lines in each functional assay.

FACS analysisECFCs were suspended in PBS and 1 x 106 cells was used for antibodies staining for 30 minutesat 4°C. The antibodies used were as follow and according to the manufacturer’s instructions:FITC-conjugated anti-CD34 (BD Pharmingen), FITC-conjugated anti-CD45 (Biolegend), PE-conjugated anti-VEGFR2 (R&D system), FITC-conjugated anti-CD31 (Miltenyi Biotech) andFITC-conjugated anti-VE-cadherin (AbD serotec). The cells were filtered using a 40 μm cellstrainer and transfer to a Falcon tube for FACSCanto (BD Pharmingen) analysis and the cellsurface markers as expressed percentages were analyzed by FlowJo.

Far infrared radiation treatmentThe cells were exposed to radiation from aWSTM TY301 FIR emitter (Far IR Medical Technol-ogy, Taipei, Taiwan). The wavelength of the light generated from the electrified ceramic platesranged from 5 to 12 μmwith a peak at 8.2 μm. The radiator was placed 30 cm above the bottomof the culture plates for the indicated times. The functional assays were preformed 24 hoursafter FIR treatment.

Small RNA sequencing (smRNA-seq) and data analysisTotal RNA was collected from one disease free ECFCs (dfECFC) and one high glucose treateddfECFCs (HG-dfECFCs). Small RNA fractions were sequenced using the Illumina GAIIx sys-tem (Illumina, San Diego, CA USA) following the manufacturer’s standard procedure.Sequencing data were analyzed by a published in-house bioinformatics pipeline [32]. The pipe-line was conducted to process mRNA contamination, to identify known miRNA quantifica-tion, and to carry out format conversion. The RPM (reads per mapped reads) value for eachmiRNA was calculated in order to score its expression abundance. SVMicrO (compgenomics.utsa.edu/svmicro.html) [33] and miRTar (mirtar.mbc.nctu.edu.tw/) were [34] were used formiRNA downstream targets prediction.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 5 / 27

Reporter assaysIn order to create the luciferase reporter plasmids by cloning, the putative miRNA-binding sitewas cloned into the XbaI site of the pGL3-Basic plasmid (Promega), while a mutated binding sitewas generated by OuickChange™ Site-Directed Mutagenesis Kit (Stratagene). The primers usedare listed in S2 Table. For detection of luciferase intensities, miRNA and reporter plasmids wereco-transfected into 293T cells using Turbofect reagent (Thermo Scientific™). Cell lysates were col-lected after 48 hours and their luciferase activities measured using a Dual-Luciferase ReporterAssay Kit (Promega). The luciferase activities were calculated based on the ratio of firefly lucifer-ase to renilla luciferase activity and further normalized against the pGL3 vector control group.

The mouse ischemic hindlimb model and ECFCs transplantationNude mice at 8 to 10 weeks of age received right femoral artery excision surgery to induce uni-lateral hindlimb ischemia as previously described [35]. The mice were anesthetized by an intra-peritoneal injection of Tribromoethanol (Avertin1) and the right femoral artery was excised.Mice were arbitrarily arranged into four groups for different treatments: EGM2 medium, HG-dfECFCs, FIR treated HG-dfECFCs and FIR treated HG-dfECFCs with miR-134 overexpres-sion. ECFCs were pre-stained with PKH26 (Sigma-Aldrich) for 30 minutes and transplanted at3 days after surgery. Next 2.5 × 105 ECFCs in 60 μl medium were injected intramuscularly atsix different sites on the ischemic limb distal to the arterial occlusion site [36]. Blood perfusionwas determined by Laser Doppler Perfusion Imager system (LDPI, Moor Instruments Limited,Devon, UK) before surgery, immediately after surgery and weekly following surgery. Bloodflow was analyzed by moor LDI software version 5.3 and the results are indicated as the ratio ofperfusion in the ischemic (right) versus non-ischemic (left) limb in each mouse. Cell countingand capillary density were analyzed using ImageJ by determining the PKH26/CD31 doublepositive cells and CD31 positive region, respectively.

Immunofluorescence stainingFrozen section slides were treated with 200 ul of 3% H2O2 for 15 minutes at room temperatureand were blocked with 5% BSA in 0.2% PBST for 1 hr. Then, slides were washed with PBS for 5minutes for 3 times with shacking. A α-mouse CD31 antibody (Santa Cruz) in 0.2% BSA-PBSTwas used for detection of capillary.

Statistical analysesThe results were statistically analyzed using the software SPSS v20. For experiments consist oftwo groups of samples, Mann-Whitney U test was used to evaluate the statistical significance.One-way ANOVA followed by Tukey’s post-hoc test was applied to determine the significanceamong each group of all double manipulation assays and other experiments with more thantwo treatment groups. A p value that is lower than 0.05 was considered statistically significant.

Results

High glucose impairs the functioning of ECFCsTo investigate the functional activities of ECFCs in DM patients, we first isolated ECFCs fromthe peripheral blood of DM patients (dmECFCs) or disease-free subjects (dfECFCs). These twotypes of ECFCs both had a cobblestone-like morphology that was similar to mature endothelialcells [3, 4] (Fig 1A). Using flow cytometry analysis, both dm- and df-ECFCs expressed the pre-cursor marker CD34 and endothelial markers, PECAM1, VE-cadherin and VEGFR2, butlacked hematopoietic lineage marker CD45 (S1A Fig). To assess their functioning, we

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 6 / 27

Fig 1. Reduced angiogenic activities in dmECFCs and high glucose treated dfECFCs. (A) The cell morphology of ECFCs from disease-free (df) anddiabetes mellitus (dm) individuals Scale bar: 50 μm. (B) Representative images (left) and quantitative data (right) for the Transwell cellular migration assay(upper) and microvascular formation assay (lower) using dfECFCs and dmECFCs. n = 3 independent experiments. ** p < 0.01 by Mann-Whitney U test.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 7 / 27

compared the Transwell migration and tube formation ability of dmECFCs to df-ECFCs andfound that dmECFCs showed 80% and 30% reduction in cell motility and the formation ofmicrovasculature structure, respectively (Fig 1B). dmECFCs also proliferated 30% slower thanthat of dfECFCs on day 3 and this difference had further increased to 55% on day 6, as mea-sured by MTT assay (Fig 1C).

In order to explore whether the impaired functioning of ECFCs was caused by high glucose,the isolated dfECFCs were treated for 72 hours with high glucose (HG) alone to mimic thehyperglycemia found in DM patients, or HG combined with low growth factor (LGF) condi-tions (HG/LGF) to mimic the poorest progression of DM patients. dfECFCs were also treatedwith D-mannitol (NG) and LGF used as control group for the HG and HG/LGF group, respec-tively. [28]. The morphologies (Fig 1D) and the expressed surface markers (S1B Fig) of all 4groups of treated cells remained similar. Comparing to the NG and LGF groups, the HG andHG/LGF groups showed 35% and 40% decreases in cell migration, and also 63% and 74%reductions in tube formation ability, respectively (Fig 1E). Using the MTT assay, the HG andHG/LGF groups had a slower cell proliferation rate on day 5, with reductions of 23% and 45%,respectively (Fig 1F). However, a similar cell cycle patterns for these groups was found usingflow cytometry, which indicates that the decreased cell proliferation was not due to cell cyclearrest (S2A Fig). Similar decreases in cell motility and tube formation caused by HG treatmentwere found for HUVECs (S2B Fig). These findings suggested that HG impairs the angiogenicactivity of ECFCs.

Small RNA sequencing analysis identified a set of miRNAs that wereincreased in HG-dfECFCsIn order to explore the mechanisms that HG reduces the angiogenic related activities ofECFCs, we focused on the differences in miRNA expression patterns between HG-dfECFCsand dfECFCs using small RNA sequencing; our results were analyzed using our in-house bioin-formatics pipeline [32]. We hypothesized that a set of miRNAs induced by HG hampered thefunctioning of dfECFCs. In summary, 448 known miRNAs were found to be differentiallyexpressed between HG-dfECFCs and dfECFCs, with 281 and 167 miRNAs being more abun-dant in HG-dfECFCs and dfECFCs, respectively (> 2 fold; Fig 2A). The top ten HG-enrichedmiRNAs are shown as a bubble chart (Fig 2B and Table 1). We validated these miRNA expres-sion levels using four DF subjects’ ECFCs incubated under NG and HG condition and foundthat miR-370, miR-183-5p and miR-134 were increased under the HG condition compared tothe NG treatment condition (Fig 2C), whereas the other seven miRNAs showed no statisticallysignificant change (S3 Fig). Only miR-370 and miR-134 were used for further experimentsbecause these two miRNAs were found to be more abundant in dmECFCs compared todfECFCs (Fig 2D).

miR-134 regulates ECFCmotility and microvasculature formation abilityWe next investigated the angiogenic functioning of miR-370 and miR-134 in dfECFCs. Afteroverexpression of these two miRNAs in dfECFCs, miR-370 and miR-134 were 11-fold and

Scale bar: 50 μm (C) Cell proliferation rates of dfECFCs and dmECFCs measured using the MTT assay. ** p < 0.01, *** p < 0.001 by Mann-Whitney U test.n = 3 independent experiments. (D) Cell morphology of dfECFCs treated with normal culture (NG), high glucose (HG), low growth factor (LGF) and HG/LGF.Scale bar: 50 μm (E) Representative images (lower) and quantitative data (upper) of Transwell migration assay and microvascular formation assay usingdfECFCs under NG, HG, LGF, and HG/LGF conditions. n = 3 independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVA followedTukey’s post-hoc test. (F) Cell proliferation rate of dfECFCs under NG, HG, LGF, HG/LGF treatment measured using the MTT assay. n = 3 independentexperiments. * p < 0.05, ** p < 0.01, by by one-way ANOVA followed by Tukey’s post-hoc test.

doi:10.1371/journal.pone.0147067.g001

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 8 / 27

3.6-fold increased respectively (S4A Fig). Cell motility and microvasculature formation werethen determined and it was found that miR-134 reduced cell migration by 34% and tube for-mation ability by 51%, while overexpression of miR-370 in dfECFCs resulted in no significantchange (Fig 3A). Both miR-370 and miR-134 had no effect on cell proliferation or on cell cyclearrest (S4B and S4C Fig). We also measured the effect of miR-370 and miR-134 overexpressionon cell motility and microvasculature formation of HUVECs (S4D Fig). The findings were sim-ilar to that of ECFCs with only miR-134 overexpression leading to a decrease in cell migrationand tube formation (S4E Fig).

Next we used an oligonucleotide antagomir to reduce the level of miR-134 in dfECFCsunder NG and HG treatment conditions (Fig 3B). The decreased level of miR-134 in HG-dfECFCs was more than that in NG-dfECFCs (a 6.8 fold decrease in HG-dfECFCs vs. a 2.4 folddecreased in NG-dfECFCs) because the original level of miR-134 was higher in HG-dfECFCs.

Fig 2. Small RNA sequencing and RT-qPCR reveal high level of expression of miR-134 in HG-dfECFCs and dmECFCs. (A) Differentially expressedmiRNAs between HG-dfECFCs and dfECFCs. Numbers of dominant miRNAs in each ECFC type (fold change > 2) are shown. Y-axis was shown the foldratio of RPM of HG-dfECFCs against dfECFCs. (B) The top ten HG-enriched miRNAs are shown. Y-axis was indicated the log2 ratio of HG-dfECFC againstdfECFCs. The diameters of the circles indicate the log2RPM of the HG-dfECFCs. (C) The putative HG-induced miRNAs were validated using dfECFCstreated with NG and HG by RT-qPCR. * p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVA followed by Tukey’s post-hoc test. (D) The HG-inducedmiRNAs were also analyzed in dmECFCs and dfECFCs by RT-qPCR. * p < 0.05, ** p < 0.01 by one-way ANOVA followed by Tukey’s post-hoc test.

doi:10.1371/journal.pone.0147067.g002

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 9 / 27

Inhibition of miR-134 did not increase cell migration and tube formation in NG-dfECFCs.However, knock-down of miR-134 partially restored the reduced functionality of HG-dfECFCs(Fig 3C and S5 Fig). In addition, ECFCs from DM patients also expressed higher level of miR-134 than DF subjects. Inhibition of miR-134 in dmECFCs at 2.7 fold (Fig 3D) improved themigration and microvasculature formation activity of these cells by 3 fold and 2 fold, respec-tively (Fig 3E). These results suggested that miR-134, which is abundant in HG-dfECFCs anddmECFCs, hampers the angiogenic activity of ECFCs.

Far infrared radiation rescues the functioning of ECFCsSeveral previous studies have reported that FIR improves vascular endothelial function andaugments angiogenesis [22, 23, 26]. We therefore explored whether FIR treatment restore thedecreased activities of HG-dfECFCs and dmECFCs. HG-dfECFCs were treated with FIR for 30or 40 minutes and the cells were then subjected to the Transwell migration and tube formationassays. Compared to HG-dfECFCs without FIR pretreatment, FIR improved HG-dfECFCmotility and microvasculature structure formation by at least 1.8 and 1.6 fold, respectively (Fig4A and S6A Fig). FIR also promoted these changes in dmECFC activities in a dose dependentmanner (Fig 4B). FIR treatment of 30 to 60 minutes increased dmECFC motility by 2-fold to3.5-fold and enhanced microvasculature formation from 1.2-fold to 1.5-fold (Fig 4B and S6BFig). We next explored whether FIR improved the activities of HG-dfECFCs and dmECFCs viaan inhibition of miR-134 expression. We detected miR-134 expression using RT-qPCR indfECFCs from four subjects treated with or without FIR and found that the level of miR-134was decreased after FIR treatment in four independent experiments (Fig 4C). We next used adouble manipulation approach to confirm the relationship between FIR and miR-134 (Fig 4D).The decreased angiogenic activity caused by miR-134 overexpression was found to be rescuedby FIR treatment (Fig 4E). In order to determine the persistent effect of FIR, HG-dfECFCs andHG-dfECFCs with FIR treatment were subjected to Transwell migration and MTT assays. Theexpression of miR-134 was reduced by FIR and was persistent from day1 to day 10 after FIRtreatment (S7A Fig). ECFC migration ability was enhanced by FIR treatment and this effectwas persistent for 10 days after FIR treatment. However, the numbers of migrated cells weregradually decreased in both cell groups because of the prolong cultivation under HG condition(S7B Fig). MTT assay performed from the indicated day after FIR treatment showed FIR slightlypromoted cell proliferation to around day 10 (S7C and S7D Fig) and then cell tend to die in HG-

Table 1. Top 10 enrichedmiRNAs in HG-dfECFCs.

RPM Fold

Name dfECFC HG-dfECFC HG-df / df

hsa-mir-122-5p 84.143196 93717.86 1113.790116

hsa-mir-370 23.864511 5048.3394 211.5417073

hsa-mir-183-5p 38.978706 5603.7925 143.7654831

hsa-mir-494 119.32256 11111.615 93.12249922

hsa-mir-182-5p 320.84512 25406.03 79.18471691

hsa-mir-134 330.12573 25924.89 78.53035266

hsa-mir-142-3p 3.8044875 262.93384 69.11150056

hsa-mir-432-5p 34.240387 1866.4335 54.50970808

hsa-mir-376a-5p 95.88805 5047.9097 52.64378304

hsa-mir-4448 3.7865787 184.72424 48.78394314

The reads of each miRNAs was normalized with total reads and indicated as reads per million (RPM). Fold is the reads of HG-dfECFC against dfECFC.

doi:10.1371/journal.pone.0147067.t001

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 10 / 27

Fig 3. miR-134 regulates ECFCmotility andmicrovasculature formation ability. (A) Representative images (left) and quantitative data (right) from theTranswell migration assays (upper) and tube formation assays (lower) of dfECFCs with miR-370 and miR-134 overexpression. n = 3 independent

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 11 / 27

containing media (S7E Fig, indicated with #). These findings suggested that FIR is able to reducemiR-134 to promote the functions of ECFCs and this effect could be persisted at least 10 days.

Microarray analyses reveal that NRIP1 was the miR-134 targetIn order to determine the mechanism involved in regulating ECFC activity, we identified genesinvolved in FIR treatment by microarray analysis (GSE37044) [37]. We focused on genes thatwere enhanced by FIR (Fig 5A) and predicted targets of miR-134 using SVMicrO. The 135 geneswere subject to miRTar to narrow down the putative targets of miR-134 and 35 putative targetswere identified (Fig 5A). Among the top 10 enhanced genes that are putative miR-134 targets, 6genes were reported to be involved in diabetes or angiogenesis. These six genes, namely potas-sium channel, inwardly rectifying subfamily J, member2 (KCNJ2) [38], cytoplasmic polyadenyla-tion element binding protein 4 (CPEB4) [39], nuclear receptor-interacting protein 1 (NRIP1),transient receptor potential cation channel, subfamily M, member 7 (TRPM7) [40, 41], Kruppel-like factor 3 (KLF3) [42] and suppressor of cytokine signaling 5 (SOCS5) [43] (Fig 5A). Thesegenes were then validated using RT-qPCR. Isolated dfECFCs from four individuals were treatedwith FIR to determine the expression of the six targets. We found that only NRIP1was increasedin all four FIR treated ECFC samples compared to non-FIR treated cells (Fig 5B and S8A Fig). Inaddition, NRIP1 showed a lower expression level in HG-dfECFCs compared to NG-dfECFCs(Fig 5C). NRIP1 was also reduced in ECFCs from clinical DM patients (Fig 5D). These resultssuggest that FIR treatment is able to reduce miR-134 expression, which in turn enhances thedownstream expression level of NRIP1. In contrast, DM or HG treatment leads to miR-134 upre-gulation in ECFCs and a decrease in NRIP1 expression.

We further investigated whether NIRP1 is a downstream target of miR-134. Overexpressionof miR-134 in dfECFCs (S8B Fig) results in a reduction of 90% in NRIP1 expression (Fig 5E).Combining FIR treatment with miR-134 overexpressed ECFCs reversed this decrease inNRIP1 expression and results in a similar level of expression to that found in the vector control(Fig 5F). Furthermore, NRIP1 expression not only significantly increased in both antagomirtreated NG-dfECFCs and HG-dfECFCs (Fig 5G and S8C Fig), but also in dmECFCs with miR-134 antagomir treatment (Fig 5H). We next used a 3’UTR luciferase reporter assay to clarifythe direct targeting of NRIP1 by miR-134. The putative miR-134 binding region within the3’UTR of NRIP1 is shown in Fig 5I. miR-134 overexpression repressed the luciferase activity ofthe wild type NRIP1 3’UTR, but this reduction was reversed using a mutated NRIP1’s 3’UTR(Fig 5J). These findings suggested that miR-134 directly targets the 3’ UTR of NRIP1 to attenu-ate angiogenic activity of ECFCs in vitro.

NRIP1 is involved in various ECFC activities and is crucial for miR-134functionalityIn order to explore the role of NRIP1 in ECFCs, we used shRNA to knockdown NRIP1 indfECFCs (Fig 6A) and this led to a 2.5-fold reduction in cell motility and a 2-fold reduction in

experiments. * p < 0.05 by one-way ANOVA followed by Tukey’s post-hoc test. Scale bar: 50 μm (B) The expression levels of miR-134 under osmotic culturecondition (NG) or under high glucose condition (HG) treated with scramble control and miR-134 antagomir (Anti-134) in dfECFCs as quantified by RT-qPCR.* p < 0.05 by one-way ANOVA followed by Tukey’s post-hoc test n = 3 independent experiments. (C) Cellular migration assays (left) and tube formationassays (right) of NG-dfECFCs and HG-dfECFCs after scramble control and miR-134 antagomir (Anti-134) treatment. n = 3 independent experiments. *p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVA followed by Tukey’s post-hoc test. (D) Expression levels of miR-134 in dfECFCs or antagomirtreated dmECFCs using RT-qPCR. n = 3 independent experiments. * p < 0.05 by one-way ANOVA followed by Tukey’s post-hoc test. (E) Representativeimages (lower) and quantitative data (upper) from the Transwell migration assays and tube formation assays of dfECFCs and antagomir treated dmECFCs.n = 3 independent experiments. * p < 0.05, ** p < 0.01 by one-way ANOVA followed by Tukey’s post-hoc test. Scale bar: 50 μm. Vec, plasmid control; Scr,scramble; Anti-134, miR-134 antagomir.

doi:10.1371/journal.pone.0147067.g003

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 12 / 27

Fig 4. FIR treatment improves angiogenic activities of ECFCs. (A)Quantitative data from the Transwell cell migration assays (left) and tube formationassays (right) using HG-dfECFCs with or without FIR treatment. n = 3 independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVAfollowed by Tukey’s post-hoc test. (B)Quantitative data from the Transwell cell migration assays (left) and tube assays (right) using dmECFCs treated withFIR. n = 3 independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001 by one-way ANOVA followed by Tukey’s post-hoc test. (C) The expression levelsof miR-134 in dfECFCs from four individuals with or without FIR treatment. * p < 0.05 by one-way ANOVA followed by Tukey’s post-hoc test. (D)miR-134

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 13 / 27

microvasculature structure formation (Fig 6B). We further used an anti-miR-134 antagomircombined with shNRIP1 to clarify the interaction between miR-134 and NRIP1 (Fig 6C). Theincrease in cell migration and tube formation activities caused by miR-134 inhibition wasreduced by NRIP1 knockdown in dfECFCs (Fig 6D). These results suggest that NRIP1 is adirect target of miR-134 and is involved in ECFCs functioning. We also determined the relationbetween FIR and NRIP1 by combining FIR treatment in NRIP1 knockdown or control HG-dfECFCs. The expression of NRIP1 was determined to clarify the knockdown efficiency (S9AFig). The FIR induced microvasculature formation in plasmid control HG-dfECFC was sup-pressed by inhibition of NRIP1 (S9B and S9C Fig). The results suggest the FIR modulates thefunctions of ECFCs through miR-134-NRIP1 axis.

FIR induces ECFC activation and promotes blood flow recovery using amouse ischemic limb modelWe next evaluated the therapeutic potential of FIR in terms of the in vivo angiogenic activity ofHG-dfECFCs. Nude mice received femoral artery excision surgery to develop limb ischemiathree days before ECFC injection. At 48 hours prior to transplantation, we overexpressed a vectorcontrol and miR-134 in the HG-dfECFCs. Furthermore, the ECFCs were treated with FIR oneday before ECFC injection (Fig 7A). At day 0 after injection, miR-134 and NRIP1 were detectedusing RT-qPCR to confirm that miR-134 expression was reduced and NRIP1 expression wasincreased by FIR (S10A Fig). A reduction in blood flow was observed in left leg after surgery (day3) until ECFC injection (day 0) (Fig 7B). HG-dfECFCs with different treatment were injectedintramuscularly at the ischemic hindlimb from distal to the arterial occlusion site 3 days after sur-gery and the blood flow was then measured by Laser Doppler perfusion imaging (LDPI) at day 7and day 14. Although blood flow in the ischemic leg had recovered slightly in the medium con-trol group (the “EGM2” group) on day 14 (Fig 7B), HG-dfECFCs transplantation (the “Vec”group) improved the blood flow ratio from 0.29 on day 0 to 0.6 on day 7 and 0.73 on day 14. Inaddition, the improvement of blood flow was more significant in FIR-treated HG-dfECFCs (the“Vec + FIR” group) with an increase in the ratio to 0.88 on day 7 and 1.0 on day 14, which isclose to the pre-operation condition. Overexpression of miR-134 in the FIR-treated HG-dfECFCs(the “miR-134 + FIR”) reduced the flow ratio to 0.67 on day 7 and 0.77 on day 14 (Fig 7B).

To investigate the effects of FIR on the homing and angiogenic activities of the injectedECFCs, we conducted immunofluorescence staining of the mouse limb tissues at day 7 (Fig7C) and 14 (S10B Fig) after HG-dfECFCs injection. Capillary density was examined as anindex of angiogenesis using immunostaining with CD31 antibody on day 7 (green, Fig 7C) andday 14 (S10B Fig). The injected human HG-dfECFCs pre-labeled with PKH-26 fluorescenceindicated that homing and involvement of ECFCs in microvascular repair did occur (red, Fig7C). Mice in the FIR treated HG-dfECFCs group (the “Vec + FIR” group) presented with moreCD31+/PKH-26+ double-positive cells (white arrowheads) in the capillaries of the limb musclethan the Vec and miR-134/FIR groups (Fig 7C; quantitative data in Fig 7D). In addition, thecapillary densities of the limb muscle were also higher in mice treated with FIR than the Vecand miR-134/FIR groups (Fig 7D). The findings obtained using the ischemia limb model sug-gested that FIR is able to restore the angiogenic activity of HG-dfECFCs and improve vascularrepair in mice in vivo.

expression in vector control and miR-134 overexpressed dfECFC with or without FIR treatment. * p < 0.05, ** p < 0.01, by one-way ANOVA followed byTukey’s post-hoc test. (E)Representative images (lower) and quantitative data (upper) from the Transwell cell migration assays and microvascular formationassays using vector control or miR-134 overexpressed dfECFCs with or without FIR treatment. n = 3 independent experiments. * p < 0.05, ** p < 0.01 byone-way ANOVA followed Tukey’s post-hoc test. Scale bar: 50 μm.

doi:10.1371/journal.pone.0147067.g004

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 14 / 27

Fig 5. NRIP1 is a direct target of miR-134. (A) A Venn diagram shows the principle used to filter the target genes of miR-134. (B, C) Validation of putativemiR-134 downstream targets in dfECFCs under FIR (B) and high glucose (C) treatment using RT-qPCR. * p < 0.05, ** p < 0.01 by one-way ANOVA followedby Tukey’s post-hoc test. (D) Relative expression level of NRIP1 in dfECFCs and dmECFCs. *** p < 0.001 by Mann-Whitney U test (E, F) Validation of

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 15 / 27

DiscussionType 2 diabetes mellitus is a chronic metabolic disease and is a growing public health issue inthe world [44]. T2D is associated with insulin resistance, hyperglycemia and dyslipidemia andis a risk factor for cardiovascular diseases [45]. In addition, patients with DM are predisposedto endothelial dysfunction and several vascular complications [46, 47]. Several mechanisms ofendothelial dysfunction in DM have been identified including alteration in signaling related toeNOS activation, increased oxidative stress, activation of inflammatory processes and impairedmitochondrial functioning [48]. In this study, we focus on the effects of high glucose on a sub-set of endothelial progenitor cells, namely ECFCs, which display clonal proliferative potentialand in vivo vascular formation activity [49–51]. Several studies have suggested that ECFCs area potential cell source for regenerative medicine because of the ability of ECFCs to home ontoischemic regions, to increase microvascular density, to promote vascular repair and to helprecovery of blood flow [52–55]. Furthermore, ECFCs with high glucose treatment or isolatedfrom GDM pregnancies have a slower cell proliferation rate, impaired cell migration abilityand reduced tube formation ability [9, 10]. In this study, ECFCs treated with HG also showsimilar results to those presented in a previous study. In addition, ECFCs treated with HG/LGFshow poorer ECFC functionalities than the HG group (Fig 1E and 1F) and this result is reason-able since the HG/LGF group mimics the physical reaction of the poorest progression of DMpatients. We have also identified that HG increases three miRNAs in ECFCs, namely miR-370,miR-183-5p and miR-134. However, only miR-370 and miR-134 show enhanced expressionlevels in dmECFCs. Overexpression of miR-134, but not of miR-370, in dfECFCs results indecreased cell migration and poorer tube formation ability (Fig 3A). Although dmECFCs showhigher level of miR-370 than dfECFCs, it seems that miR-370 may also regulate other cellularmechanisms in DM. According to previous studies, miR-370 is a lipid metabolism-relatedmiRNA [56] and the increased miR-370 expression in hyperlipidemia patients has been posi-tively correlated with the severity of coronary artery disease [57]. Several genes, including ste-rol-regulatory element binding protein 1c (SREBP-1c), enzymes diacylglycerol acyltransferase-2 (DGAT2), fatty acid synthase (FAS) and acyl-CoA carboxylase 1 (ACC1), that are allinvolved in regulation of fatty acid and triglyceride biosynthesis, are increased by miR-370 inHepG2 cells. miR-370 also targets carnitine palmitoyltransferase 1α (Cpt1α), which is a mito-chondria enzyme and is involved in transportation of long-chain fatty acid for energy produc-tion, resulting in a decreased rate of beta oxidation [58]. The accumulation of hepatic fatty acidhas been shown to be associated with insulin resistance and metabolic syndrome.

Previous studies have suggested that miR-134 overexpression promotes neuronal cell deathand apoptosis by regulating response element-binding protein (CREB) signaling and targetingheat shock protein A12B (HSPA12B) [59, 60]. In retinal ganglion cells (RGCs), miR-134 isincreased by H2O2 treatment and lead to RGC apoptosis [61]. However, miR-134 overexpres-sion in ECFCs did not alter their cell proliferation rate based on the results from the MTTassays or result in cell cycle arrest based on the flow cytometry analysis. Our results and previ-ous studies both show that decreased ECFC proliferation in DM and in high glucose condition,

NRIP1 expression in miR-134 overexpressed dfECFCs (E) and in combination with FIR treatment (F). n = 3 independent experiments. * p < 0.05 (E) byMann-Whitney U test, * p < 0.05, ** p < 0.01 (F) by one-way ANOVA followed by Tukey’s post-hoc test. (G, H) The expression levels of NRIP1 in HG-dfECFCs or dmECFCs treated with scramble control or miR-134 antagomir. n = 3 independent experiments. * p < 0.05, ** p < 0.01 one-way ANOVAfollowed by Tukey’s post-hoc test. (I) Structure of the NRIP1 transcript and the predicted miR-134 binding site on the NRIP1-3’UTR. (J) The relative luciferaseactivities of the vector control or miR-134 overexpressed 293T cells co-transfected with the wild type or mutant NRIP1 3’UTR reporter plasmids (right panels).Luciferase activity was normalized against the vector control group and are presented as the mean + SD. The expression levels of each miRNAs as detectedby RT-qPCR (left panels). n = 3 independent experiments. * p < 0.05 by one-way ANOVA followed by Tukey’s post-hoc test. Scr, scramble control; Vec,plasmid control; Wt, wild type 3’UTR of NRIP1; Mut, mutated 3’UTR of NRIP1.

doi:10.1371/journal.pone.0147067.g005

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 16 / 27

Fig 6. NRIP1 is involved in ECFC activity and is crucial for miR-134 functionality. (A) Validation of NRIP1 in dfECFCs infected with NRIP1 shRNA.* p < 0.05 by Mann-Whitney U test. (B) Representative images (left) and quantitative data (right) from the Transwell migration assays (upper) and

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 17 / 27

but it seems that this occurs in a miR-134-independent manner. In contrast, miR-134 was firstidentified as an angiogenic modulator that decreases ECFC migration and microvascular for-mation. Hyperglycemia damages endothelial cells by several mechanisms including increasedreactive oxygen species, decreased nitric oxide (NO) bioavailability, augmentation of glycationend products (AGEs) and an over-activated hexosamine pathway [62]. HG-treated ECFCshave been shown to have decreased levels of phosphorylated endothelial nitric oxide synthase(eNOS) and bioavailable NO [9]. However, whether hyperglycemia also impairs ECFCs viathese mechanisms in endothelial cells is still unknown. Furthermore, whether reactive nitrogenspecies (RNS) or reactive oxygen species (ROS) participate in the HG-induced miR-134 modu-lation of ECFCs motility and tube formation need further investigation.

We explored how NRIP1 regulates ECFC activities and is the target of miR-134 in thisstudy. NRIP1 controls several physiological responses via binding with other transcription fac-tors and histone modifying enzymes such that it can act as either a coactivator or a corepressor[63]. A previous study has reported that NRIP1 expression is higher in diabetes patients andcorrelates with inflammatory cytokines in plasma and peripheral blood mononuclear cells(PBMCs) [64]. In human skeletal muscle, insulin triggers the mitogen-activated protein kinase(MAPK) and phosphoinositide 3-kinase (PI3K)-Akt pathways to phosphorylate TBC1 domainfamily, member 4 (TBC1D4, also called AS160), a Rab-GTPase-activating protein (Rab-GAP)[65]. This then results in the translocation of glucose transporter type 4 (GLUT4) storage vesi-cles to plasma membrane and increased glucose uptake. However, cytoplasmic NRIP1 impairsGLUT4 trafficking via interacting with AS160, which reduces the level of Akt-mediated phos-phorylation [66]. These results suggest that NRIP1 acts as negative regulator in relation to glu-cose uptake and thus would seem likely to contribute to various metabolism diseases. Furtherinvestigation is needed to address how NRIP1 modulates ECFC activities and if any factorsthat bind with NRIP1 are able to exert angiogenic activities.

FIR is an electromagnetic radiation with wavelengths from 5.6–1000 um. FIR has beenapplied to a variety of fields such as food preservation [67] and health improvement, includingincreased growth, better sleep [20], better microcirculation [25, 68], improved wound healing[69] and decreased chronic pain [21]. In this study, FIR treated HG-dfECFCs improved bloodflow recovery in hindlimb ischemia model, which is a simpler model of peripheral arterial dis-ease and is useful for testing new therapies. Notably, the variability in the result of Laser Dopp-ler results of blood flow may arise from inconsistently positioned leg and different sizes regionsof interest. We minimized the variability in data analysis by fixing selected size in interestedregion with moor LDI software. A previous study has shown that FIR activates p38 and extra-cellular signal-regulated kinase (ERK), rather than Akt and c-Jun N-terminal protein kinases(JNK), to promote angiogenic activities after 30 minutes of FIR radiation [70]. In renal cell car-cinoma cells, miR-134 also modulates the MAPK/ERK pathway and acts as a tumor suppressor[71]. However, whether the above-mentioned signaling pathways are affected by miR-134 reg-ulation under FIR treatment and the detailed interactions that occur between miR-134 andthese pathways need further investigation. Nevertheless, miR-134 seems to be a marker fordetermining the effectiveness of FIR treatment.

Endothelial nitric oxide synthase (eNOS), an enzyme that generates nitric oxide (NO), iscrucial for normal endothelial functions, including modulation of vascular dilator tone and

microvascular formation assays (lower) using ECFCs infected with control or NRIP1 shRNA plasmids. n = 3 independent experiments. * p < 0.05 by Mann-Whitney U test. Scale bar: 50 μm. (C) The expression levels of miR-134 and NRIP1 in dfECFCs treated with miR-134 antagomir and NRIP1 shRNA as adouble manipulation. n = 3 independent experiments. * p < 0.05, ** p < 0.01 by one-way ANOVA followed by Tukey’s post-hoc test. (D) Representativeimages (upper) and quantitative data (lower) of the Transwell cell migration assays (left) and tube formation assays (right) using dfECFCs with miR-134antagomir and NRIP1 shRNA as a double manipulation. n = 3 independent experiments. * p < 0.05, ** p < 0.01 by one-way ANOVA followed by Tukey’spost-hoc test. Scale bar: 50 μm. Ctrl: plasmid control, shNRIP1: NRIP1 shRNA, Anti-miR-134: miR-134 antagomir.

doi:10.1371/journal.pone.0147067.g006

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 18 / 27

Fig 7. FIR induces ECFC activation and improves blood perfusion using an ischemic hindlimbmodel. (A) Schematic representation of experimentaldesign. (B) Upper: quantitative analysis of blood flow expressed as perfusion ratio of the ischemic to the non-operated contralateral hindlimb. n = 6,** p < 0.01 by one-way ANOVA followed by Tukey’s post-hoc test. Lower: representative images of mouse ventral side during the measurement of hindlimb

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 19 / 27

protection of vessels from injurious consequences. Dysregulation of eNOS activities and dimin-ish release of NO contribute to several physical diseases. Repeat FIR sauna treatment increaseseNOS, augments angiogenesis [26, 72] and attenuates cardiac remodeling after myocardialinfarction (MI) [73]. However, HMVECs treated with FIR emitter do not show an increase inphosphorylated eNOS. Based on our microarray datasets, there is also no significant enhance-ment of eNOS in ECFCs treated with FIR when comparing to non-FIR treatment. The discrep-ancy may be due to differences in the FIR devices used or to differences in the duration of FIRtreatment; these differences also need further investigation.

In addition to hyperglycemia, elevated free fatty acids (FFA) and dysregulated hormonesare common issues in diabetes and contribute to endothelial dysfunctions. Comparing theobese nondiabetic subjects with T2D patients, saturated fatty acids (SFA), but not polyunsatu-rated fatty acids (PUFA), increase significantly in the plasma of DM patients [74]. Several stud-ies have showed that FFAs decrease NOS activity and reduce NO production in endothelialcells by impairing PLC-medicated Ca2+ signaling or inhibiting insulin-induced insulin receptorsubstrate 1 (IRS-1) and Akt pathway [75, 76]. Saturated fatty acid, especially palmitic acid(PA), is the most studied fatty acid in ECFCs. PA activates p38 MAPK, JNK or Akt to increaseapoptosis or decrease angiogenic activities of ECFCs in a dose-dependent manner [77, 78].However, the effects of PA are different depend on the concentration. Only diabetic PA, butnot physiologic PA, impairs ECFC proliferation through decreasing phosphorylation form ofSTAT5 [79]. Whether other saturated, monounsaturated or polyunsaturated FAs, regulateECFC angiogenic activities remains unclear.

Stress response is a reaction to deal with anything that disrupts the body homeostasis.Chronic stress induces hormones, such as glucocorticoids (GC) and catecholamines, which notonly regulate angiogenic activities [80, 81], but also lead to various manifestations of metabolicsyndrome, including diabetes [82]. ECFCs isolated from GC-induced avascular osteonecrosisof the femoral head show decreased cell proliferation and tube formation capacities when com-pared to healthy subjects [83]. In addition, dopamine induces endocytosis of VEGF receptor 2and results in preventing VEGFA-induced angiogenesis [84]. Dopamine also reduces VEGFA-induced mobilization of bone marrow CD45-CD34+VEGFR2+ cells to peripheral blood andresults in restrained tumor growth [85]. However, norepinephrine shows opposite effect onmobilization of CD45-CD34+VEGFR2+ cells in mice with hindlimb ischemia [86]. The conse-quence of FFA and stress-induced hormones to ECFCs and mature endothelial cells varies con-siderably depending on the cellular milieu and need further studies.

ConclusionsIn this study, we are the first to provide the evidence that FIR treatment suppresses miR-134,which is induced under HG condition. The reduction of miR-134 results in increasing NRIP1level and thus improves the angiogenic activities of ECFCs in vitro. The FIR treated HG-dfECFCs also show that this treatment has in vivo therapeutic potential since it improves bloodflow perfusion using a hindlimb ischemic model. Our findings provide a biological effect forFIR and suggest that FIR treatment may be able to improve the progression of DM patients.

blood flow by laser Doppler before operation (Pre-Op), immediately after hindlimb ischemia surgery (Post-Op), and 2 weeks after intramuscular injection ofculture medium (EGM2), HG-dfECFCs (Vec), HG-dfECFCs with FIR treatment (Vec + FIR), and miR-134 overexpressed HG-dfECFCs with FIR treatment(miR-134 + FIR). (C) Immunofluorescence staining of the tissue from nude mice 7 days after injection with PKH-26-labeled HG-dfECFCs. The capillaries inthe limb muscles were visualized by anti-CD31 immunostaining (green), and injected human ECFCs were monitored by PKH-26 fluorescence (red). Inaddition there was Hoechst nuclear staining of the live cells (blue). (D)Quantitative analysis of CD31+/PKH-26+ double-positive cells and capillary densitiesin limb muscle of mice hindlimb ischemia region. HPF: high power field; N.D.: not detectable; * p < 0.05 by one-way ANOVA followed by Tukey’s post-hoctest.

doi:10.1371/journal.pone.0147067.g007

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 20 / 27

Supporting InformationS1 Fig. Expression of indicated molecules in ECFCs using flow cytometry analyses. (A)FACS analysis of the surface antigens present on dfECFCs and dmECFCs. (B) The cell surfacemarkers expressed on NG-dfECFCs, HG-dfECFCs, LGF-dfECFCs and HG/LGF-dfECFCs.(TIF)

S2 Fig. The cell cycle of ECFCs and the angiogenic activities of HUVECs under NG, HG,LGF and HG/LGF conditions. (A) Cell cycle analysis of dfECFCs under normal culture (NG),high glucose (HG), low growth factor (LGF) and HG/LGF conditions using flow cytometry.(B) Representative images (upper) and quantitative data (lower) for the cell migration assaysand tube formation assays using HUVECs. �� p< 0.01, ��� p< 0.001 by one-way ANOVA fol-lowed by Tukey’s post-hoc test. Scale bar: 50 μm.(TIF)

S3 Fig. RT-qPCR results showing the expression levels of 7 miRNAs present in microarraysof NG-dfECFCs and HG-dfECFCs. � p< 0.05, �� p< 0.01 by one-way ANOVA followed byTukey’s post-hoc test.(TIF)

S4 Fig. The effect of miR-370 and miR-134 overexpression on dfECFCs or HUVECs. (A)dfECFCs were infected with lenti-miR-370 and lenti-miR-134. The expression levels ofmiRNA are quantified by RT-qPCR. �� p< 0.01 by one-way ANOVA followed by Tukey’spost-hoc test. (B) Cell proliferation rate in dfECFCs with miR-370 and miR-134 overexpres-sion. (C) Flow cytometry analysis of the cell cycles of miR-370 and miR-134 overexpresseddfECFCs. (D)Overexpression of miR-370 and miR-134 in HUVECs and the expression levelsof these miRNAs as quantified by RT-qPCR. � p< 0.05, �� p< 0.01 by one-way ANOVA fol-lowed by Tukey’s post-hoc test. (E) Representative images (lower) and quantitative data(upper) for the Transwell migration assays and tube formation assays using HUVECs withmiR-370 and miR-134 overexpression. � p< 0.05, �� p< 0.01 by one-way ANOVA followedby Tukey’s post-hoc test. Scale bar: 50 μm.(TIF)

S5 Fig. Representative images of the cell migration assays and tube formation assays ofscramble control or miR-134 antagomir treated NG-dfECFCs and HG-dfECFCs. Scale bar:50 μm.(TIF)

S6 Fig. Representative images of the cell migration assays and tube formation assays ofHG-dfECFCs (A) and dmECFCs (B) after FIR treatment. Scale bar: 50 μm.(TIF)

S7 Fig. The persistent effect of FIR. (A) The expression of miR-134 at indicated days afterFIR treatment in HG-dfECFCs. � p< 0.05, �� p< 0.01 by one-way ANOVA followed byTukey’s post-hoc test. (B) The quantative data of migrated cell numbers at indicated days afterFIR treatment. � p< 0.05, �� p< 0.01 by one-way ANOVA followed by Tukey’s post-hoc test.(C~E)MTT cell proliferation assay performed after 3, 8 and 10 days after FIR treatment.� p< 0.05 by Mann-Whitney U test.(TIF)

S8 Fig. Validation of the putative downstream targets of miR-134. (A) Validation of theputative miR-134 target genes by RT-qPCR in dfECFCs with or without FIR treatment.� p< 0.05, �� p< 0.01, ��� p< 0.001 by one-way ANOVA followed by Tukey’s post-hoc test.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 21 / 27

(B, C) The expression level of miR-134 in miR-134 overexpressed ECFCs (B) and miR-134antagomir treated ECFCs (C). � p< 0.05 (B) by Mann-Whitney U test, �� p< 0.01, ��� p<0.001 (C) by one-way ANOVA followed by Tukey’s post-hoc test.(TIF)

S9 Fig. Tube formation ability in NRIP1 knockdown HG-dfECFCs or plasmid control HG-dfECFCs either combined FIR treatment. (A) The expression levels of NRIP1 in each groupof ECFCs. � p< 0.05, �� p< 0.01 by one-way ANOVA followed by Tukey’s post-hoc test. (B)The quantitative data of tube formation assay. � p< 0.05 by one-way ANOVA followed byTukey’s post-hoc test. (C) Representative images of the microvascular formation assays in eachgroup cells. Scale bar: 50 μm.(TIF)

S10 Fig. Transplantation of FIR treated HG-dfECFCs improves blood perfusion in ische-mic hindlimb regions. (A) Validation of the expression levels of miR-134 and NRIP1 in eachgroup of ECFCs by RT-qPCR on day 0. � p< 0.05 by one-way ANOVA followed by Tukey’spost-hoc test. (B) Immunofluorescence staining of nude mice tissue samples on day 14 afterinjection with HG-dfECFCs. Capillaries in the ischemic muscles were visualized by anti-CD31immunostaining (green), Hoechst: nuclear staining of live cells (blue).(TIF)

S1 Table. qPCR Primer Sequences.(XLSX)

S2 Table. Cloning Primer Sequences.(XLSX)

AcknowledgmentsThis work was supported by Tri-Service General Hospital [TSGH-C103-023, TSGH-C104-028], Ministry of Science and Technology [MOST; 103-2321-B-010-026, 103-2911-I-010-506,104-2911-I-010-506 and 104-2321-B-101-027], Veterans General Hospitals University Systemof Taiwan (VGHUST) Joint Research Program, Tsou’s Foundation [VGHUST103-G7-2-3],National Health Research Institutes [NHRI-EX103-10254SI], Taipei City Hospital [10401-62-014], National Science Council [102-2911-I-010-503, 101-2320-B-010-059-MY3] and NationalYang-Ming University [Ministry of Education, Aim for the Top University Plan]. This workwas also support in part by the UST-UCSD International Center for Excellence in AdvancedBioengineering sponsored by the Taiwan MOST I-RiCE Program [103-2911-I-009-101]. Theauthors would like to dedicate this paper to the memory of Dr. Hsei-Wei Wang, who passedaway during of this research. Without his long-term devotion to endothelial cell research, thispaper could not have been completed.

Author ContributionsConceived and designed the experiments: HWWCCC. Performed the experiments: SHS YLWKHL HHL YLC THH. Analyzed the data: SHS YLW STC TSH. Contributed reagents/materi-als/analysis tools: CSL SMC CCC. Wrote the paper: SHS HWW.

References1. Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetes: estimates for the year

2000 and projections for 2030. Diabetes care. 2004; 27(5):1047–53. PMID: 15111519

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 22 / 27

2. Martin A, Komada MR, Sane DC. Abnormal angiogenesis in diabetes mellitus. Medicinal researchreviews. 2003; 23(2):117–45. PMID: 12500286

3. Hur J, Yoon CH, Kim HS, Choi JH, Kang HJ, Hwang KK, et al. Characterization of two types of endothe-lial progenitor cells and their different contributions to neovasculogenesis. Arteriosclerosis, thrombosis,and vascular biology. 2004; 24(2):288–93. PMID: 14699017

4. Ingram DA, Mead LE, Tanaka H, Meade V, Fenoglio A, Mortell K, et al. Identification of a novel hierar-chy of endothelial progenitor cells using human peripheral and umbilical cord blood. Blood. 2004; 104(9):2752–60. PMID: 15226175

5. Mukai N, Akahori T, Komaki M, Li Q, Kanayasu-Toyoda T, Ishii-Watabe A, et al. A comparison of thetube forming potentials of early and late endothelial progenitor cells. Experimental cell research. 2008;314(3):430–40. PMID: 18083163

6. Medina RJ, O'Neill CL, Humphreys MW, Gardiner TA, Stitt AW. Outgrowth endothelial cells: characteri-zation and their potential for reversing ischemic retinopathy. Investigative ophthalmology & visual sci-ence. 2010; 51(11):5906–13.

7. He T, Smith LA, Harrington S, Nath KA, Caplice NM, Katusic ZS. Transplantation of circulating endothe-lial progenitor cells restores endothelial function of denuded rabbit carotid arteries. Stroke; a journal ofcerebral circulation. 2004; 35(10):2378–84. PMID: 15345801

8. Dubois C, Liu X, Claus P, MarsboomG, Pokreisz P, Vandenwijngaert S, et al. Differential effects of pro-genitor cell populations on left ventricular remodeling and myocardial neovascularization after myocar-dial infarction. Journal of the American College of Cardiology. 2010; 55(20):2232–43. doi: 10.1016/j.jacc.2009.10.081 PMID: 20466204

9. Chen YH, Lin SJ, Lin FY, Wu TC, Tsao CR, Huang PH, et al. High glucose impairs early and late endo-thelial progenitor cells by modifying nitric oxide-related but not oxidative stress-mediated mechanisms.Diabetes. 2007; 56(6):1559–68. PMID: 17389326

10. Gui J, Rohrbach A, Borns K, Hillemanns P, Feng L, Hubel CA, et al. Vitamin D rescues dysfunction offetal endothelial colony forming cells from individuals with gestational diabetes. Placenta. 2015; 36(4):410–8. doi: 10.1016/j.placenta.2015.01.195 PMID: 25684656

11. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004; 116(2):281–97.PMID: 14744438

12. Chang TC, Mendell JT. microRNAs in vertebrate physiology and human disease. Annual review ofgenomics and human genetics. 2007; 8:215–39. PMID: 17506656

13. Esquela-Kerscher A, Slack FJ. Oncomirs—microRNAs with a role in cancer. Nature reviews Cancer.2006; 6(4):259–69. PMID: 16557279

14. Bushati N, Cohen SM. MicroRNAs in neurodegeneration. Current opinion in neurobiology. 2008; 18(3):292–6. doi: 10.1016/j.conb.2008.07.001 PMID: 18662781

15. van Rooij E, Olson EN. MicroRNAs: powerful new regulators of heart disease and provocative thera-peutic targets. The Journal of clinical investigation. 2007; 117(9):2369–76. PMID: 17786230

16. Guay C, Roggli E, Nesca V, Jacovetti C, Regazzi R. Diabetes mellitus, a microRNA-related disease?Translational research: the journal of laboratory and clinical medicine. 2011; 157(4):253–64.

17. Hu S, Huang M, Li Z, Jia F, Ghosh Z, Lijkwan MA, et al. MicroRNA-210 as a novel therapy for treatmentof ischemic heart disease. Circulation. 2010; 122(11 Suppl):S124–31. doi: 10.1161/CIRCULATIONAHA.109.928424 PMID: 20837903

18. Grundmann S, Hans FP, Kinniry S, Heinke J, Helbing T, Bluhm F, et al. MicroRNA-100 regulates neo-vascularization by suppression of mammalian target of rapamycin in endothelial and vascular smoothmuscle cells. Circulation. 2011; 123(9):999–1009. doi: 10.1161/CIRCULATIONAHA.110.000323PMID: 21339483

19. Vatansever F, Hamblin MR. Far infrared radiation (FIR): its biological effects and medical applications.Photonics & lasers in medicine. 2012; 4:255–66.

20. Honda K, Inoue S. Sleep-enhancing effects of far-infrared radiation in rats. International journal ofbiometeorology. 1988; 32(2):92–4. PMID: 3410585

21. Masuda A, Koga Y, Hattanmaru M, Minagoe S, Tei C. The effects of repeated thermal therapy forpatients with chronic pain. Psychotherapy and psychosomatics. 2005; 74(5):288–94. PMID: 16088266

22. Imamura M, Biro S, Kihara T, Yoshifuku S, Takasaki K, Otsuji Y, et al. Repeated thermal therapyimproves impaired vascular endothelial function in patients with coronary risk factors. Journal of theAmerican College of Cardiology. 2001; 38(4):1083–8. PMID: 11583886

23. Kihara T, Biro S, Imamura M, Yoshifuku S, Takasaki K, Ikeda Y, et al. Repeated sauna treatmentimproves vascular endothelial and cardiac function in patients with chronic heart failure. Journal of theAmerican College of Cardiology. 2002; 39(5):754–9. PMID: 11869837

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 23 / 27

24. Beever R. Far-infrared saunas for treatment of cardiovascular risk factors: summary of published evi-dence. Canadian family physician Medecin de famille canadien. 2009; 55(7):691–6. PMID: 19602651

25. Yu SY, Chiu JH, Yang SD, Hsu YC, Lui WY, Wu CW. Biological effect of far-infrared therapy on increas-ing skin microcirculation in rats. Photodermatology, photoimmunology & photomedicine. 2006; 22(2):78–86.

26. Akasaki Y, Miyata M, Eto H, Shirasawa T, Hamada N, Ikeda Y, et al. Repeated thermal therapy up-reg-ulates endothelial nitric oxide synthase and augments angiogenesis in a mouse model of hindlimbischemia. Circulation journal: official journal of the Japanese Circulation Society. 2006; 70(4):463–70.

27. Wang HW, Lo HH, Chiu YL, Chang SJ, Huang PH, Liao KH, et al. Dysregulated miR-361-5p/VEGF axisin the plasma and endothelial progenitor cells of patients with coronary artery disease. PloS one. 2014;9(5):e98070. doi: 10.1371/journal.pone.0098070 PMID: 24865854

28. Caporali A, Meloni M, Vollenkle C, Bonci D, Sala-Newby GB, Addis R, et al. Deregulation of microRNA-503 contributes to diabetes mellitus-induced impairment of endothelial function and reparative angio-genesis after limb ischemia. Circulation. 2011; 123(3):282–91. doi: 10.1161/CIRCULATIONAHA.110.952325 PMID: 21220732

29. Cheng CC, Lo HH, Huang TS, Cheng YC, Chang ST, Chang SJ, et al. Genetic module and miRNometrait analyses reflect the distinct biological features of endothelial progenitor cells from different ana-tomic locations. BMC genomics. 2012; 13:447. doi: 10.1186/1471-2164-13-447 PMID: 22943456

30. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT, et al. Real-time quantification of micro-RNAs by stem-loop RT-PCR. Nucleic acids research. 2005; 33(20):e179. PMID: 16314309

31. Alaiti MA, IshikawaM, Masuda H, Simon DI, Jain MK, Asahara T, et al. Up-regulation of miR-210 byvascular endothelial growth factor in ex vivo expanded CD34+ cells enhances cell-mediated angiogen-esis. Journal of cellular and molecular medicine. 2012; 16(10):2413–21. doi: 10.1111/j.1582-4934.2012.01557.x PMID: 22360314

32. ChengWC, Chung IF, Huang TS, Chang ST, Sun HJ, Tsai CF, et al. YM500: a small RNA sequencing(smRNA-seq) database for microRNA research. Nucleic acids research. 2013; 41(Database issue):D285–94. doi: 10.1093/nar/gks1238 PMID: 23203880

33. Liu H, Yue D, Chen Y, Gao SJ, Huang Y. Improving performance of mammalian microRNA target pre-diction. BMC Bioinformatics. 2010; 11:476. doi: 10.1186/1471-2105-11-476 PMID: 20860840

34. Hsu JB, Chiu CM, Hsu SD, HuangWY, Chien CH, Lee TY, et al. miRTar: an integrated system for iden-tifying miRNA-target interactions in human. BMC bioinformatics. 2011; 12:300. doi: 10.1186/1471-2105-12-300 PMID: 21791068

35. Huang PH, Chen CY, Lin CP, Wang CH, Tsai HY, Lo WY, et al. Deletion of FHL2 gene impaired ische-mia-induced blood flow recovery by modulating circulating proangiogenic cells. Arteriosclerosis, throm-bosis, and vascular biology. 2013; 33(4):709–17. doi: 10.1161/ATVBAHA.112.300318 PMID:23413425

36. Huang PH, Chen YH, Wang CH, Chen JS, Tsai HY, Lin FY, et al. Matrix metalloproteinase-9 is essen-tial for ischemia-induced neovascularization by modulating bone marrow-derived endothelial progenitorcells. Arteriosclerosis, thrombosis, and vascular biology. 2009; 29(8):1179–84. doi: 10.1161/ATVBAHA.109.189175 PMID: 19461050

37. Wang HW, Huang TS, Lo HH, Huang PH, Lin CC, Chang SJ, et al. Deficiency of the microRNA-31-microRNA-720 pathway in the plasma and endothelial progenitor cells from patients with coronaryartery disease. Arteriosclerosis, thrombosis, and vascular biology. 2014; 34(4):857–69. doi: 10.1161/ATVBAHA.113.303001 PMID: 24558106

38. Salem KA, Qureshi MA, Sydorenko V, Parekh K, Jayaprakash P, Iqbal T, et al. Effects of exercise train-ing on excitation-contraction coupling and related mRNA expression in hearts of Goto-Kakizaki type 2diabetic rats. Molecular and cellular biochemistry. 2013; 380(1–2):83–96. doi: 10.1007/s11010-013-1662-2 PMID: 23620341

39. Ortiz-Zapater E, Pineda D, Martinez-Bosch N, Fernandez-Miranda G, Iglesias M, Alameda F, et al. Keycontribution of CPEB4-mediated translational control to cancer progression. Nature medicine. 2012; 18(1):83–90.

40. Sun H, Leng T, Zeng Z, Gao X, Inoue K, Xiong ZG. Role of TRPM7 channels in hyperglycemia-medi-ated injury of vascular endothelial cells. PloS one. 2013; 8(11):e79540. doi: 10.1371/journal.pone.0079540 PMID: 24223965

41. Zeng Z, Inoue K, Sun H, Leng T, Feng X, Zhu L, et al. TRPM7 regulates vascular endothelial cell adhe-sion and tube formation. American journal of physiology Cell physiology. 2015; 308(4):C308–18. doi:10.1152/ajpcell.00275.2013 PMID: 25472964

42. Bell-Anderson KS, Funnell AP, Williams H, Mat Jusoh H, Scully T, LimWF, et al. Loss of Kruppel-likefactor 3 (KLF3/BKLF) leads to upregulation of the insulin-sensitizing factor adipolin (FAM132A/CTRP12/C1qdc2). Diabetes. 2013; 62(8):2728–37. doi: 10.2337/db12-1745 PMID: 23633521

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 24 / 27

43. Zhuang G, Wu X, Jiang Z, Kasman I, Yao J, Guan Y, et al. Tumour-secreted miR-9 promotes endothe-lial cell migration and angiogenesis by activating the JAK-STAT pathway. The EMBO journal. 2012; 31(17):3513–23. doi: 10.1038/emboj.2012.183 PMID: 22773185

44. Narayan KM, Boyle JP, Geiss LS, Saaddine JB, Thompson TJ. Impact of recent increase in incidenceon future diabetes burden: U.S., 2005–2050. Diabetes care. 2006; 29(9):2114–6. PMID: 16936162

45. Eckel RH, Wassef M, Chait A, Sobel B, Barrett E, King G, et al. Prevention Conference VI: Diabetesand Cardiovascular Disease: Writing Group II: pathogenesis of atherosclerosis in diabetes. Circulation.2002; 105(18):e138–43. PMID: 11994264

46. Golden SH. Emerging therapeutic approaches for the management of diabetes mellitus and macrovas-cular complications. The American journal of cardiology. 2011; 108(3 Suppl):59B–67B. doi: 10.1016/j.amjcard.2011.03.017 PMID: 21802582

47. Edwards MS, Wilson DB, Craven TE, Stafford J, Fried LF, Wong TY, et al. Associations between retinalmicrovascular abnormalities and declining renal function in the elderly population: the CardiovascularHealth Study. American journal of kidney diseases: the official journal of the National Kidney Founda-tion. 2005; 46(2):214–24.

48. Kolluru GK, Bir SC, Kevil CG. Endothelial dysfunction and diabetes: effects on angiogenesis, vascularremodeling, and wound healing. Int J Vasc Med. 2012; 2012:918267. doi: 10.1155/2012/918267 PMID:22611498

49. Yoder MC, Mead LE, Prater D, Krier TR, Mroueh KN, Li F, et al. Redefining endothelial progenitor cellsvia clonal analysis and hematopoietic stem/progenitor cell principals. Blood. 2007; 109(5):1801–9.PMID: 17053059

50. Ingram DA, Mead LE, Moore DB, WoodardW, Fenoglio A, Yoder MC. Vessel wall-derived endothelialcells rapidly proliferate because they contain a complete hierarchy of endothelial progenitor cells.Blood. 2005; 105(7):2783–6. PMID: 15585655

51. Critser PJ, Kreger ST, Voytik-Harbin SL, Yoder MC. Collagen matrix physical properties modulateendothelial colony forming cell-derived vessels in vivo. Microvasc Res. 2010; 80(1):23–30. doi: 10.1016/j.mvr.2010.03.001 PMID: 20219180

52. Schwarz TM, Leicht SF, Radic T, Rodriguez-Araboalaza I, Hermann PC, Berger F, et al. Vascular incor-poration of endothelial colony-forming cells is essential for functional recovery of murine ischemic tis-sue following cell therapy. Arteriosclerosis, thrombosis, and vascular biology. 2012; 32(2):e13–21. doi:10.1161/ATVBAHA.111.239822 PMID: 22199368

53. Heo SC, Kwon YW, Jang IH, Jeong GO, Yoon JW, Kim CD, et al. WKYMVm-induced activation of for-myl peptide receptor 2 stimulates ischemic neovasculogenesis by promoting homing of endothelial col-ony-forming cells. Stem cells. 2014; 32(3):779–90. doi: 10.1002/stem.1578 PMID: 24155208

54. Moubarik C, Guillet B, Youssef B, Codaccioni JL, Piercecchi MD, Sabatier F, et al. Transplanted lateoutgrowth endothelial progenitor cells as cell therapy product for stroke. Stem cell reviews. 2011; 7(1):208–20. doi: 10.1007/s12015-010-9157-y PMID: 20526754

55. Huang XT, Zhang YQ, Li SJ, Li SH, Tang Q, Wang ZT, et al. Intracerebroventricular transplantation ofex vivo expanded endothelial colony-forming cells restores blood-brain barrier integrity and promotesangiogenesis of mice with traumatic brain injury. Journal of neurotrauma. 2013; 30(24):2080–8. doi: 10.1089/neu.2013.2996 PMID: 23957220

56. Fernandez-Hernando C, Suarez Y, Rayner KJ, Moore KJ. MicroRNAs in lipid metabolism. Current opin-ion in lipidology. 2011; 22(2):86–92. doi: 10.1097/MOL.0b013e3283428d9d PMID: 21178770

57. GaoW, He HW,Wang ZM, Zhao H, Lian XQ, Wang YS, et al. Plasma levels of lipometabolism-relatedmiR-122 and miR-370 are increased in patients with hyperlipidemia and associated with coronaryartery disease. Lipids in health and disease. 2012; 11:55. doi: 10.1186/1476-511X-11-55 PMID:22587332

58. Iliopoulos D, Drosatos K, Hiyama Y, Goldberg IJ, Zannis VI. MicroRNA-370 controls the expression ofmicroRNA-122 and Cpt1alpha and affects lipid metabolism. Journal of lipid research. 2010; 51(6):1513–23. doi: 10.1194/jlr.M004812 PMID: 20124555

59. HuangW, Liu X, Cao J, Meng F, Li M, Chen B, et al. miR-134 regulates ischemia/reperfusion injury-induced neuronal cell death by regulating CREB signaling. Journal of molecular neuroscience: MN.2015; 55(4):821–9. doi: 10.1007/s12031-014-0434-0 PMID: 25316150

60. Chi W, Meng F, Li Y, Li P, Wang G, Cheng H, et al. Impact of microRNA-134 on neural cell survivalagainst ischemic injury in primary cultured neuronal cells and mouse brain with ischemic stroke by tar-geting HSPA12B. Brain research. 2014; 1592:22–33. doi: 10.1016/j.brainres.2014.09.072 PMID:25304362

61. Shao Y, Yu Y, Zhou Q, Li C, Yang L, Pei CG. Inhibition of miR-134 Protects Against Hydrogen Perox-ide-Induced Apoptosis in Retinal Ganglion Cells. Journal of molecular neuroscience: MN. 2015; 56(2):461–71. doi: 10.1007/s12031-015-0522-9 PMID: 25744098

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 25 / 27

62. Sena CM, Pereira AM, Seica R. Endothelial dysfunction—a major mediator of diabetic vascular dis-ease. Biochim Biophys Acta. 2013; 1832(12):2216–31. doi: 10.1016/j.bbadis.2013.08.006 PMID:23994612

63. Nautiyal J, Christian M, Parker MG. Distinct functions for RIP140 in development, inflammation, andmetabolism. Trends in endocrinology and metabolism: TEM. 2013; 24(9):451–9. doi: 10.1016/j.tem.2013.05.001 PMID: 23742741

64. Xue J, Zhao H, Shang G, Zou R, Dai Z, Zhou D, et al. RIP140 is associated with subclinical inflamma-tion in type 2 diabetic patients. Experimental and clinical endocrinology & diabetes: official journal, Ger-man Society of Endocrinology [and] German Diabetes Association. 2013; 121(1):37–42.

65. Karlsson HK, Zierath JR, Kane S, Krook A, Lienhard GE, Wallberg-Henriksson H. Insulin-stimulatedphosphorylation of the Akt substrate AS160 is impaired in skeletal muscle of type 2 diabetic subjects.Diabetes. 2005; 54(6):1692–7. PMID: 15919790

66. Ho PC, Lin YW, Tsui YC, Gupta P, Wei LN. A negative regulatory pathway of GLUT4 trafficking in adi-pocyte: new function of RIP140 in the cytoplasm via AS160. Cell metabolism. 2009; 10(6):516–23. doi:10.1016/j.cmet.2009.09.012 PMID: 19945409

67. Lloyd BJ, Farkas BE, Keener KM. Characterization of radiant emitters used in food processing. TheJournal of microwave power and electromagnetic energy: a publication of the International MicrowavePower Institute. 2003; 38(4):213–24.

68. Lin CC, Chang CF, Lai MY, Chen TW, Lee PC, YangWC. Far-infrared therapy: a novel treatment toimprove access blood flow and unassisted patency of arteriovenous fistula in hemodialysis patients.Journal of the American Society of Nephrology: JASN. 2007; 18(3):985–92. PMID: 17267744

69. Toyokawa H, Matsui Y, Uhara J, Tsuchiya H, Teshima S, Nakanishi H, et al. Promotive effects of far-infrared ray on full-thickness skin wound healing in rats. Experimental biology and medicine. 2003; 228(6):724–9. PMID: 12773705

70. Rau CS, Yang JC, Jeng SF, Chen YC, Lin CJ, Wu CJ, et al. Far-infrared radiation promotes angiogene-sis in human microvascular endothelial cells via extracellular signal-regulated kinase activation. Photo-chemistry and photobiology. 2011; 87(2):441–6. doi: 10.1111/j.1751-1097.2010.00853.x PMID:21143604

71. Liu Y, Zhang M, Qian J, Bao M, Meng X, Zhang S, et al. miR-134 Functions as a Tumor Suppressor inCell Proliferation and Epithelial-to-Mesenchymal Transition by Targeting KRAS in Renal Cell Carci-noma Cells. DNA and cell biology. 2015; 34(6):429–36. doi: 10.1089/dna.2014.2629 PMID: 25811077

72. Huang PH, Chen JW, Lin CP, Chen YH, Wang CH, Leu HB, et al. Far infra-red therapy promotes ische-mia-induced angiogenesis in diabetic mice and restores high glucose-suppressed endothelial progeni-tor cell functions. Cardiovascular diabetology. 2012; 11:99. doi: 10.1186/1475-2840-11-99 PMID:22894755

73. Sobajima M, Nozawa T, Shida T, Ohori T, Suzuki T, Matsuki A, et al. Repeated sauna therapy attenu-ates ventricular remodeling after myocardial infarction in rats by increasing coronary vascularity of non-infarcted myocardium. American journal of physiology Heart and circulatory physiology. 2011; 301(2):H548–54. doi: 10.1152/ajpheart.00103.2011 PMID: 21622828

74. Clore JN, Allred J, White D, Li J, Stillman J. The role of plasma fatty acid composition in endogenousglucose production in patients with type 2 diabetes mellitus. Metabolism. 2002; 51(11):1471–7. PMID:12404200

75. Tang Y, Li G. Chronic exposure to high fatty acids impedes receptor agonist-induced nitric oxide pro-duction and increments of cytosolic Ca2+ levels in endothelial cells. Journal of molecular endocrinol-ogy. 2011; 47(3):315–26. doi: 10.1530/JME-11-0082 PMID: 21994216

76. Kim F, Tysseling KA, Rice J, PhamM, Haji L, Gallis BM, et al. Free fatty acid impairment of nitric oxideproduction in endothelial cells is mediated by IKKbeta. Arteriosclerosis, thrombosis, and vascular biol-ogy. 2005; 25(5):989–94. PMID: 15731493

77. Jiang H, Liang C, Liu X, Jiang Q, He Z, Wu J, et al. Palmitic acid promotes endothelial progenitor cellsapoptosis via p38 and JNKmitogen-activated protein kinase pathways. Atherosclerosis. 2010; 210(1):71–7. doi: 10.1016/j.atherosclerosis.2009.10.032 PMID: 20226460

78. GuoWX, Yang QD, Liu YH, Xie XY, Wang M, Niu RC. Palmitic and linoleic acids impair endothelial pro-genitor cells by inhibition of Akt/eNOS pathway. Arch Med Res. 2008; 39(4):434–42. doi: 10.1016/j.arcmed.2008.02.001 PMID: 18375256

79. Trombetta A, Togliatto G, Rosso A, Dentelli P, Olgasi C, Cotogni P, et al. Increase of palmitic acid con-centration impairs endothelial progenitor cell and bone marrow-derived progenitor cell bioavailability:role of the STAT5/PPARgamma transcriptional complex. Diabetes. 2013; 62(4):1245–57. doi: 10.2337/db12-0646 PMID: 23223023

80. Small GR, Hadoke PW, Sharif I, Dover AR, Armour D, Kenyon CJ, et al. Preventing local regenerationof glucocorticoids by 11beta-hydroxysteroid dehydrogenase type 1 enhances angiogenesis.

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 26 / 27

Proceedings of the National Academy of Sciences of the United States of America. 2005; 102(34):12165–70. PMID: 16093320

81. Seya Y, Fukuda T, Isobe K, Kawakami Y, Takekoshi K. Effect of norepinephrine on RhoA, MAP kinase,proliferation and VEGF expression in human umbilical vein endothelial cells. European journal of phar-macology. 2006; 553(1–3):54–60. PMID: 17070516

82. Kyrou I, Tsigos C. Stress hormones: physiological stress and regulation of metabolism. Current opinionin pharmacology. 2009; 9(6):787–93. doi: 10.1016/j.coph.2009.08.007 PMID: 19758844

83. Chen C, Yang S, Feng Y, Wu X, Chen D, Yu Q, et al. Impairment of two types of circulating endothelialprogenitor cells in patients with glucocorticoid-induced avascular osteonecrosis of the femoral head.Joint, bone, spine: revue du rhumatisme. 2013; 80(1):70–6.

84. Basu S, Nagy JA, Pal S, Vasile E, Eckelhoefer IA, Bliss VS, et al. The neurotransmitter dopamine inhib-its angiogenesis induced by vascular permeability factor/vascular endothelial growth factor. Naturemedicine. 2001; 7(5):569–74. PMID: 11329058

85. Chakroborty D, Chowdhury UR, Sarkar C, Baral R, Dasgupta PS, Basu S. Dopamine regulates endo-thelial progenitor cell mobilization frommouse bone marrow in tumor vascularization. The Journal ofclinical investigation. 2008; 118(4):1380–9. doi: 10.1172/JCI33125 PMID: 18340382

86. Jiang Q, Ding S, Wu J, Liu X, Wu Z. Norepinephrine stimulates mobilization of endothelial progenitorcells after limb ischemia. PloS one. 2014; 9(7):e101774. doi: 10.1371/journal.pone.0101774 PMID:25007164

miR-134 and Endothelial Cell Dysfunction

PLOS ONE | DOI:10.1371/journal.pone.0147067 January 22, 2016 27 / 27